1 |
#!/usr/bin/perl -w |
2 |
|
3 |
use strict; |
4 |
|
5 |
use Cwd qw/abs_path/; |
6 |
use File::Temp qw/tempdir/; |
7 |
use Data::Dumper; |
8 |
use lib './lib'; |
9 |
|
10 |
use WebPAC::Common 0.02; |
11 |
use WebPAC::Lookup; |
12 |
use WebPAC::Input 0.03; |
13 |
use WebPAC::Store 0.03; |
14 |
use WebPAC::Normalize::XML; |
15 |
use WebPAC::Output::TT; |
16 |
use WebPAC::Output::Estraier 0.05; |
17 |
use YAML qw/LoadFile/; |
18 |
use LWP::Simple; |
19 |
|
20 |
my $limit = shift @ARGV; |
21 |
|
22 |
my $config = LoadFile('conf/config.yml'); |
23 |
|
24 |
print "config = ",Dumper($config); |
25 |
|
26 |
die "no databases in config file!\n" unless ($config->{databases}); |
27 |
|
28 |
my $total_rows = 0; |
29 |
|
30 |
while (my ($database, $db_config) = each %{ $config->{databases} }) { |
31 |
|
32 |
my $log = _new WebPAC::Common()->_get_logger(); |
33 |
|
34 |
# |
35 |
# open Hyper Estraier database |
36 |
# |
37 |
|
38 |
my $est_config = $config->{hyperestraier} || $log->logdie("can't find 'hyperestraier' part in confguration"); |
39 |
$est_config->{database} = $database; |
40 |
|
41 |
$log->info("using HyperEstraier URL $est_config->{masterurl}"); |
42 |
|
43 |
my $est = new WebPAC::Output::Estraier( |
44 |
%{ $est_config }, |
45 |
); |
46 |
|
47 |
# |
48 |
# now WebPAC::Store |
49 |
# |
50 |
my $abs_path = abs_path($0); |
51 |
$abs_path =~ s#/[^/]*$#/#; |
52 |
|
53 |
my $db_path = $config->{webpac}->{db_path} . '/' . $database; |
54 |
|
55 |
$log->info("working on $database in $db_path"); |
56 |
|
57 |
my $db = new WebPAC::Store( |
58 |
path => $db_path, |
59 |
database => $database, |
60 |
debug => 1, |
61 |
); |
62 |
|
63 |
|
64 |
# |
65 |
# now, iterate through input formats |
66 |
# |
67 |
|
68 |
my @inputs; |
69 |
if (ref($db_config->{input}) eq 'ARRAY') { |
70 |
@inputs = @{ $db_config->{input} }; |
71 |
} elsif ($db_config->{input}) { |
72 |
push @inputs, $db_config->{input}; |
73 |
} else { |
74 |
$log->info("database $database doesn't have inputs defined"); |
75 |
} |
76 |
|
77 |
my @supported_inputs = keys %{ $config->{webpac}->{inputs} }; |
78 |
|
79 |
foreach my $input (@inputs) { |
80 |
|
81 |
my $type = lc($input->{type}); |
82 |
|
83 |
die "I know only how to handle input types ", join(",", @supported_inputs), " not '$type'!\n" unless (grep(/$type/, @supported_inputs)); |
84 |
|
85 |
my $lookup = new WebPAC::Lookup( |
86 |
lookup_file => $input->{lookup}, |
87 |
); |
88 |
|
89 |
my $input_module = $config->{webpac}->{inputs}->{$type}; |
90 |
|
91 |
$log->info("working on input $input->{path} [$input->{type}] using $input_module"); |
92 |
|
93 |
my $input_db = new WebPAC::Input( |
94 |
module => $input_module, |
95 |
code_page => $config->{webpac}->{webpac_encoding}, |
96 |
limit => $input->{limit}, |
97 |
lookup => $lookup, |
98 |
); |
99 |
$log->logdie("can't create input using $input_module") unless ($input); |
100 |
|
101 |
my $maxmfn = $input_db->open( |
102 |
path => $input->{path}, |
103 |
code_page => $input->{encoding}, # database encoding |
104 |
); |
105 |
|
106 |
my $n = new WebPAC::Normalize::XML( |
107 |
# filter => { 'foo' => sub { shift } }, |
108 |
db => $db, |
109 |
lookup_regex => $lookup->regex, |
110 |
lookup => $lookup, |
111 |
prefix => $input->{name}, |
112 |
); |
113 |
|
114 |
my $normalize_path = $input->{normalize}->{path}; |
115 |
|
116 |
if ($normalize_path =~ m/\.xml$/i) { |
117 |
$n->open( |
118 |
tag => $input->{normalize}->{tag}, |
119 |
xml_file => $input->{normalize}->{path}, |
120 |
); |
121 |
} elsif ($normalize_path =~ m/\.(?:yml|yaml)$/i) { |
122 |
$n->open_yaml( |
123 |
path => $normalize_path, |
124 |
tag => $input->{normalize}->{tag}, |
125 |
); |
126 |
} |
127 |
|
128 |
foreach my $pos ( 0 ... $input_db->size ) { |
129 |
|
130 |
my $row = $input_db->fetch || next; |
131 |
|
132 |
my $mfn = $row->{'000'}->[0]; |
133 |
|
134 |
if (! $mfn || $mfn !~ m#^\d+$#) { |
135 |
$log->warn("record $pos doesn't have valid MFN but '$mfn', using $pos"); |
136 |
$mfn = $pos; |
137 |
push @{ $row->{'000'} }, $pos; |
138 |
} |
139 |
|
140 |
my $ds = $n->data_structure($row); |
141 |
|
142 |
$est->add( |
143 |
id => $input->{name} . "/" . $mfn, |
144 |
ds => $ds, |
145 |
type => $config->{hyperestraier}->{type}, |
146 |
); |
147 |
|
148 |
$total_rows++; |
149 |
} |
150 |
|
151 |
}; |
152 |
|
153 |
$log->info("$total_rows records indexed"); |
154 |
|
155 |
# |
156 |
# add Hyper Estraier links to other databases |
157 |
# |
158 |
if (ref($db_config->{links}) eq 'ARRAY') { |
159 |
foreach my $link (@{ $db_config->{links} }) { |
160 |
$log->info("adding link $database -> $link->{to} [$link->{credit}]"); |
161 |
$est->add_link( |
162 |
from => $database, |
163 |
to => $link->{to}, |
164 |
credit => $link->{credit}, |
165 |
); |
166 |
} |
167 |
} |
168 |
|
169 |
} |
170 |
|